Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Energy-based Latent Aligner for Incremental Learning

About

Deep learning models tend to forget their earlier knowledge while incrementally learning new tasks. This behavior emerges because the parameter updates optimized for the new tasks may not align well with the updates suitable for older tasks. The resulting latent representation mismatch causes forgetting. In this work, we propose ELI: Energy-based Latent Aligner for Incremental Learning, which first learns an energy manifold for the latent representations such that previous task latents will have low energy and the current task latents have high energy values. This learned manifold is used to counter the representational shift that happens during incremental learning. The implicit regularization that is offered by our proposed methodology can be used as a plug-and-play module in existing incremental learning methodologies. We validate this through extensive evaluation on CIFAR-100, ImageNet subset, ImageNet 1k and Pascal VOC datasets. We observe consistent improvement when ELI is added to three prominent methodologies in class-incremental learning, across multiple incremental settings. Further, when added to the state-of-the-art incremental object detector, ELI provides over 5% improvement in detection accuracy, corroborating its effectiveness and complementary advantage to existing art.

K J Joseph, Salman Khan, Fahad Shahbaz Khan, Rao Muhammad Anwer, Vineeth N Balasubramanian• 2022

Related benchmarks

TaskDatasetResultRank
Time Series OOD GeneralizationEMG
Accuracy 152.27
18
Time Series OOD GeneralizationOpportunity
S181.69
18
Time Series OOD GeneralizationUCIHAR
OOD Performance Metric 191.07
18
Time Series OOD GeneralizationUCIHAR, UniMiB-SHAR, EMG, Opportunity Aggregated
Average Performance62.37
18
Time Series OOD GeneralizationUniMiB-SHAR
OOD Result 1 Score35.63
18
Human Activity RecognitionUniMiB-SHAR
ECE0.21
5
Human Activity RecognitionEMG
ECE0.21
5
Human Activity RecognitionOpportunity
ECE12
5
Human Activity RecognitionUCIHAR
ECE0.43
5
Showing 9 of 9 rows

Other info

Follow for update